Overview

• OpenAI plans to release an open-weight language model featuring advanced reasoning capabilities, marking its first such release since GPT-2 in 2019.
• The model will allow developers to fine-tune and locally deploy it, but access to training data and full source code will remain restricted.
• Developer feedback sessions are scheduled to begin in San Francisco in the coming weeks, followed by events in Europe and the Asia-Pacific regions.
• The decision reflects OpenAI's response to competition from Meta's Llama models and DeepSeek's R1, as well as internal debates about openness in AI development.
• OpenAI recently secured $40 billion in funding, including $30 billion from SoftBank, to support its transition to a for-profit entity and ongoing AI advancements.